Main Article Content
From dawn of social networking sites, there has been an instantaneous growth in its popularity. It is because of this reason several issues has emerged. This has allure the cyber criminals. An illicit use of publishing online messages or blogs for illicit intent has become an issue of great concern. This has abundantly assist to a ever precarious threat to social networks. Therefore, user identification is necessary.
Biometric information is of two types scilicet biological and behavior. The writeprint falls in behavioral type of biometric information. This information is used for further analysis and detection purpose. This paper is regarding improvement on previously proposed framework designed for detecting user identification for social networking websites using a writeprint approach. In writeprint approach the semantic, stylometric, sentimental properties from written text of the user are captured, which is further used to detect user identity. Initially, the content written on social networking websites is downloaded. Later from the downloaded text, numerous features are extracted using text mining or web mining techniques. Most of these features are extracted natural language processing tool. Machine Learning techniques like Multi layered perceptron mode, Support Vector Machine (SVM), ensemble modeling methods in the distributed environment which is a supervised machine learning technique is used for user identity detection.